![](https://zareason.com/wp-content/uploads/2024/07/IA-Canalys-estudo-vendas-PCs-2025-Img4-912x569-1.jpg)
Nowadays, the main point of computer market marketing is to emphasize the possibilities offered by a neural processing unit (NPU)a way to accelerate hardware to handle AI tasks locally. Both AMD How much Intel already have CPUs with this feature. In the video card segment, the idea of having a portion of the hardware dedicated to handling resources that are directly related to neural networks has been around for even longer, see NVIDIA with the Tensor Core of your RTX.
O site TechPowerUp conducted a survey asking whether its readers would be willing to pay more for hardware that promises some advantage over the use of generative AI, thanks to the presence of an NPU. 26,698 votes were counted. The overwhelming majority said they would not want to pay more for it:
- 84.1% (22,455 votes) said they do not want to pay more for it;
- 8.6.6% (2,285 votes) said they were not sure;
- 7.3% (1,958 votes) said they would be willing to pay more for these new functions.
The comments on the topic also indicate that many people still do not consider AI to be something truly crucial in their routines, completely ruling out the possibility of paying more for a CPU that promises more performance for local generative AI processing. Even companies that are investing in NPU solutions are coming up against definitions imposed by the market itself.
Microsoft says that a notebook capable of natively handling AI must achieve at least 40 TOPs (trillion operations per second). Several processors with embedded NPUs are below that. Currently, this capacity is only achieved in Snapdragon X Plus processors, available in some notebooks.
By meeting the 40 TOPs required by Microsoft, the user will be able to use, for example, a Paint feature that transforms simple sketches into highly detailed images. This type of feature is present in some AI services that run in the cloud.
AI PC will drive the computer market
According to a survey by consulting firm Canalys, computers capable of handling AI natively will represent around 40% of total sales in 2025, surpassing the mark of 100 million computers that have this feature. This represents quite a leap, considering the 18% of sales projected for 2024. This growth is also directly linked to an average increase in the price of products:
In 2028, taking into account an average annual growth of 44%, there will be 205 million PCs sold that aim to use AI resources.
Source: https://www.hardware.com.br/noticias/pesquisa-revela-que-usuarios-nao-querem-pagar-mais-por-hardware-com-recursos-de-ia.html